koishi-plugin-xhs-parser 0.1.3 → 0.1.4

package/README.md

@@ -45,6 +45,8 @@ plugins:
     showVideo: true
     downloadVideoAsFile: false
     videoDownloadMode: buffer
+    maxDownloadedVideoSizeMB: 20
+    maxVideoSendSizeMB: 100
 ```
 
 说明：
@@ -54,5 +56,8 @@ plugins:
 - `videoDownloadMode: buffer`：以二进制 Buffer 传给 `h.video()`，通常最推荐。
 - `videoDownloadMode: base64`：转换成 `base64://...` 发送，适合部分 OneBot 实现。
 - `videoDownloadMode: file`：写入系统临时目录并以 `file://` URL 发送，适合部分 Napcat / 本地网关场景。
+- `maxDownloadedVideoSizeMB: 20`：下载后的视频超过该大小时，自动回退为发送小红书视频直链，避免大视频转 base64 时超过 Node 字符串长度限制。设为 `0` 表示不限制。
+- `maxVideoSendSizeMB: 100`：视频超过该大小时不发送视频，包括直链。设为 `0` 表示不限制。
+- `loggerinfo: true`：输出视频下载开始、下载结果、大小阈值判断、发送模式和失败原因等调试日志。
 
 下载模式只处理每个笔记的首个视频，避免一次消息消耗过多带宽。

package/lib/index.d.ts

@@ -20,6 +20,8 @@ export interface Config {
     showVideo: boolean;
     downloadVideoAsFile: boolean;
     videoDownloadMode: 'buffer' | 'file' | 'base64';
+    maxDownloadedVideoSizeMB: number;
+    maxVideoSendSizeMB: number;
     showStats: boolean;
     showLink: boolean;
     showError: boolean;

package/lib/index.js

@@ -13,6 +13,7 @@ const node_fs_1 = require("node:fs");
 const parser_1 = require("./parser");
 exports.name = 'xhs-parser';
 const logger = new koishi_1.Logger(exports.name);
+const VIDEO_TOO_LARGE_MESSAGE = '[视频文件过大，跳过解析]';
 exports.Config = koishi_1.Schema.intersect([
     koishi_1.Schema.object({
         enabled: koishi_1.Schema.boolean().default(true).description('开启小红书链接/卡片解析。'),
@@ -52,6 +53,8 @@ exports.Config = koishi_1.Schema.intersect([
             koishi_1.Schema.const('base64').description('使用 base64:// 段发送视频（OneBot 常用格式，buffer 失败时可尝试）'),
             koishi_1.Schema.const('file').description('写入临时文件并通过 file:// URL 发送（Napcat 等特殊环境）'),
         ]).default('buffer').description('下载视频后的发送方式。'),
+        maxDownloadedVideoSizeMB: koishi_1.Schema.number().min(0).max(2048).step(1).default(20).description('下载视频大小上限，单位 MB。超过后自动回退为发送视频直链；设为 0 表示不限制。'),
+        maxVideoSendSizeMB: koishi_1.Schema.number().min(0).max(2048).step(1).default(100).description('视频发送大小上限，单位 MB。超过后不发送视频，包括直链；设为 0 表示不限制。'),
         showStats: koishi_1.Schema.boolean().default(true).description('展示点赞、收藏、评论、分享数据。'),
         showLink: koishi_1.Schema.boolean().default(true).description('展示原文链接。'),
     }).description('内容设置'),
@@ -136,12 +139,41 @@ async function handleLinks(ctx, session, links, config) {
     }
 }
 async function prepareNoteVideo(ctx, note, config) {
-    if (!config.downloadVideoAsFile || !config.showVideo)
+    if (!config.showVideo) {
+        if (config.loggerinfo) {
+            logger.info('skip video: showVideo=false');
+        }
         return note;
+    }
     const firstVideo = note.videoUrls[0];
-    if (!firstVideo || !/^https?:\/\//i.test(firstVideo))
+    if (!firstVideo) {
+        if (config.loggerinfo)
+            logger.info('skip video download: no video URL found');
+        return note;
+    }
+    if (!/^https?:\/\//i.test(firstVideo)) {
+        if (config.loggerinfo)
+            logger.info(`skip video download: unsupported video URL protocol (${firstVideo})`);
         return note;
+    }
+    const maxSendSizeBytes = getSizeLimitBytes(config.maxVideoSendSizeMB, 100);
+    const remoteSize = await getRemoteVideoSize(ctx, firstVideo, config);
+    if (remoteSize !== undefined && maxSendSizeBytes && remoteSize > maxSendSizeBytes) {
+        if (config.loggerinfo) {
+            logger.info(`skip video send: remote size=${formatBytes(remoteSize)}, max=${formatBytes(maxSendSizeBytes)}, url=${firstVideo}`);
+        }
+        return removeNoteVideos(note);
+    }
+    if (!config.downloadVideoAsFile) {
+        if (config.loggerinfo) {
+            logger.info('skip video download: downloadVideoAsFile=false, use direct URL');
+        }
+        return note;
+    }
     try {
+        if (config.loggerinfo) {
+            logger.info(`download first video start: url=${firstVideo}`);
+        }
         const file = await ctx.http.file(firstVideo);
         if (!file?.data) {
             if (config.loggerinfo)
@@ -151,11 +183,29 @@ async function prepareNoteVideo(ctx, note, config) {
         const buffer = Buffer.from(file.data);
         const mimeType = file.type || file.mime || 'video/mp4';
         const mode = config.videoDownloadMode || 'buffer';
+        const maxSizeBytes = config.maxDownloadedVideoSizeMB > 0
+            ? config.maxDownloadedVideoSizeMB * 1024 * 1024
+            : 0;
         if (config.loggerinfo) {
-            logger.info(`downloaded first video (${Math.round(buffer.length / 1024)} KB), send mode: ${mode}`);
+            logger.info(`download first video success: size=${formatBytes(buffer.length)}, mime=${mimeType}, mode=${mode}, max=${maxSizeBytes ? formatBytes(maxSizeBytes) : 'unlimited'}`);
+        }
+        if (maxSendSizeBytes && buffer.length > maxSendSizeBytes) {
+            if (config.loggerinfo) {
+                logger.info(`skip video send: downloaded size=${formatBytes(buffer.length)}, max=${formatBytes(maxSendSizeBytes)}`);
+            }
+            return removeNoteVideos(note);
+        }
+        if (maxSizeBytes && buffer.length > maxSizeBytes) {
+            if (config.loggerinfo) {
+                logger.info(`downloaded video exceeds limit: size=${formatBytes(buffer.length)}, max=${formatBytes(maxSizeBytes)}, fallback=direct URL`);
+            }
+            return note;
         }
         const remainingVideoUrls = note.videoUrls.slice(1);
         if (mode === 'buffer') {
+            if (config.loggerinfo) {
+                logger.info(`use downloaded video buffer: size=${formatBytes(buffer.length)}, remainingVideoUrls=${remainingVideoUrls.length}`);
+            }
             return {
                 ...note,
                 videoBuffer: buffer,
@@ -166,6 +216,9 @@ async function prepareNoteVideo(ctx, note, config) {
         const replacement = mode === 'file'
             ? await createTempVideoFile(buffer, mimeType)
             : `base64://${buffer.toString('base64')}`;
+        if (config.loggerinfo) {
+            logger.info(`use downloaded video ${mode}: src=${mode === 'file' ? replacement : `base64://${formatBytes(buffer.length)} raw`}, remainingVideoUrls=${remainingVideoUrls.length}`);
+        }
         return {
             ...note,
             videoBuffer: undefined,
@@ -179,6 +232,55 @@ async function prepareNoteVideo(ctx, note, config) {
         return note;
     }
 }
+async function getRemoteVideoSize(ctx, url, config) {
+    const maxSendSizeBytes = getSizeLimitBytes(config.maxVideoSendSizeMB, 100);
+    if (!maxSendSizeBytes)
+        return undefined;
+    try {
+        if (config.loggerinfo)
+            logger.info(`check remote video size start: url=${url}`);
+        const headers = await ctx.http.head(url, { timeout: config.timeout * 1000 });
+        const contentLength = headers.get('content-length');
+        if (!contentLength) {
+            if (config.loggerinfo)
+                logger.info('check remote video size skipped: missing content-length');
+            return undefined;
+        }
+        const size = Number.parseInt(contentLength, 10);
+        if (!Number.isFinite(size) || size < 0) {
+            if (config.loggerinfo)
+                logger.info(`check remote video size skipped: invalid content-length=${contentLength}`);
+            return undefined;
+        }
+        if (config.loggerinfo)
+            logger.info(`check remote video size success: size=${formatBytes(size)}, max=${formatBytes(maxSendSizeBytes)}`);
+        return size;
+    }
+    catch (error) {
+        if (config.loggerinfo)
+            logger.info(`check remote video size failed: ${error instanceof Error ? error.message : String(error)}`);
+        return undefined;
+    }
+}
+function removeNoteVideos(note) {
+    return {
+        ...note,
+        videoBuffer: undefined,
+        videoUrls: [],
+        videoSkippedMessage: VIDEO_TOO_LARGE_MESSAGE,
+    };
+}
+function getSizeLimitBytes(sizeMB, fallbackMB) {
+    const normalized = sizeMB ?? fallbackMB;
+    return normalized > 0 ? normalized * 1024 * 1024 : 0;
+}
+function formatBytes(bytes) {
+    if (bytes < 1024)
+        return `${bytes} B`;
+    if (bytes < 1024 * 1024)
+        return `${(bytes / 1024).toFixed(1)} KB`;
+    return `${(bytes / 1024 / 1024).toFixed(1)} MB`;
+}
 async function createTempVideoFile(buffer, mimeType) {
     const fileName = `xhs-video-${Date.now()}-${Math.random().toString(16).slice(2)}${getVideoFileExtension(mimeType)}`;
     const filePath = node_path_1.default.join(node_os_1.default.tmpdir(), fileName);

package/lib/parser.d.ts

@@ -28,6 +28,7 @@ export interface XhsNote {
     videoUrls: string[];
     videoBuffer?: Buffer;
     videoMimeType?: string;
+    videoSkippedMessage?: string;
 }
 export declare function extractXhsLinks(content: string): string[];
 export declare function resolveXhsLink(rawUrl: string, config: XhsConfigLike): Promise<string>;

package/lib/parser.js

@@ -83,6 +83,12 @@ function buildXhsMessages(note, config, session) {
             }, koishi_1.h.video(videoUrl)));
         }
     }
+    if (note.videoSkippedMessage) {
+        messages.push((0, koishi_1.h)('message', {
+            userId: session?.userId,
+            nickname: session?.author?.nickname || session?.username,
+        }, koishi_1.h.text(note.videoSkippedMessage)));
+    }
     return messages;
 }
 function extractInitialStateNote(html) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "koishi-plugin-xhs-parser",
-  "version": "0.1.3",
+  "version": "0.1.4",
   "description": "Parse Xiaohongshu links and cards for Koishi.",
   "main": "lib/index.js",
   "typings": "lib/index.d.ts",

package/src/index.ts

@@ -8,6 +8,7 @@ import { buildXhsMessages, extractXhsLinks, fetchXhsNote, XhsNote } from './pars
 export const name = 'xhs-parser'
 
 const logger = new Logger(name)
+const VIDEO_TOO_LARGE_MESSAGE = '[视频文件过大，跳过解析]'
 
 export interface Config {
   enabled: boolean
@@ -29,6 +30,8 @@ export interface Config {
   showVideo: boolean
   downloadVideoAsFile: boolean
   videoDownloadMode: 'buffer' | 'file' | 'base64'
+  maxDownloadedVideoSizeMB: number
+  maxVideoSendSizeMB: number
   showStats: boolean
   showLink: boolean
   showError: boolean
@@ -74,6 +77,8 @@ export const Config: Schema<Config> = Schema.intersect([
       Schema.const('base64').description('使用 base64:// 段发送视频（OneBot 常用格式，buffer 失败时可尝试）'),
       Schema.const('file').description('写入临时文件并通过 file:// URL 发送（Napcat 等特殊环境）'),
     ]).default('buffer').description('下载视频后的发送方式。'),
+    maxDownloadedVideoSizeMB: Schema.number().min(0).max(2048).step(1).default(20).description('下载视频大小上限，单位 MB。超过后自动回退为发送视频直链；设为 0 表示不限制。'),
+    maxVideoSendSizeMB: Schema.number().min(0).max(2048).step(1).default(100).description('视频发送大小上限，单位 MB。超过后不发送视频，包括直链；设为 0 表示不限制。'),
     showStats: Schema.boolean().default(true).description('展示点赞、收藏、评论、分享数据。'),
     showLink: Schema.boolean().default(true).description('展示原文链接。'),
   }).description('内容设置'),
@@ -164,12 +169,44 @@ async function handleLinks(ctx: Context, session: any, links: string[], config:
 }
 
 async function prepareNoteVideo(ctx: Context, note: XhsNote, config: Config): Promise<XhsNote> {
-  if (!config.downloadVideoAsFile || !config.showVideo) return note
+  if (!config.showVideo) {
+    if (config.loggerinfo) {
+      logger.info('skip video: showVideo=false')
+    }
+    return note
+  }
 
   const firstVideo = note.videoUrls[0]
-  if (!firstVideo || !/^https?:\/\//i.test(firstVideo)) return note
+  if (!firstVideo) {
+    if (config.loggerinfo) logger.info('skip video download: no video URL found')
+    return note
+  }
+  if (!/^https?:\/\//i.test(firstVideo)) {
+    if (config.loggerinfo) logger.info(`skip video download: unsupported video URL protocol (${firstVideo})`)
+    return note
+  }
+
+  const maxSendSizeBytes = getSizeLimitBytes(config.maxVideoSendSizeMB, 100)
+  const remoteSize = await getRemoteVideoSize(ctx, firstVideo, config)
+  if (remoteSize !== undefined && maxSendSizeBytes && remoteSize > maxSendSizeBytes) {
+    if (config.loggerinfo) {
+      logger.info(`skip video send: remote size=${formatBytes(remoteSize)}, max=${formatBytes(maxSendSizeBytes)}, url=${firstVideo}`)
+    }
+    return removeNoteVideos(note)
+  }
+
+  if (!config.downloadVideoAsFile) {
+    if (config.loggerinfo) {
+      logger.info('skip video download: downloadVideoAsFile=false, use direct URL')
+    }
+    return note
+  }
 
   try {
+    if (config.loggerinfo) {
+      logger.info(`download first video start: url=${firstVideo}`)
+    }
+
     const file = await ctx.http.file(firstVideo)
     if (!file?.data) {
       if (config.loggerinfo) logger.info('download video failed: empty response data')
@@ -179,14 +216,34 @@ async function prepareNoteVideo(ctx: Context, note: XhsNote, config: Config): Pr
     const buffer = Buffer.from(file.data)
     const mimeType = (file as any).type || (file as any).mime || 'video/mp4'
     const mode = config.videoDownloadMode || 'buffer'
+    const maxSizeBytes = config.maxDownloadedVideoSizeMB > 0
+      ? config.maxDownloadedVideoSizeMB * 1024 * 1024
+      : 0
 
     if (config.loggerinfo) {
-      logger.info(`downloaded first video (${Math.round(buffer.length / 1024)} KB), send mode: ${mode}`)
+      logger.info(`download first video success: size=${formatBytes(buffer.length)}, mime=${mimeType}, mode=${mode}, max=${maxSizeBytes ? formatBytes(maxSizeBytes) : 'unlimited'}`)
+    }
+
+    if (maxSendSizeBytes && buffer.length > maxSendSizeBytes) {
+      if (config.loggerinfo) {
+        logger.info(`skip video send: downloaded size=${formatBytes(buffer.length)}, max=${formatBytes(maxSendSizeBytes)}`)
+      }
+      return removeNoteVideos(note)
+    }
+
+    if (maxSizeBytes && buffer.length > maxSizeBytes) {
+      if (config.loggerinfo) {
+        logger.info(`downloaded video exceeds limit: size=${formatBytes(buffer.length)}, max=${formatBytes(maxSizeBytes)}, fallback=direct URL`)
+      }
+      return note
     }
 
     const remainingVideoUrls = note.videoUrls.slice(1)
 
     if (mode === 'buffer') {
+      if (config.loggerinfo) {
+        logger.info(`use downloaded video buffer: size=${formatBytes(buffer.length)}, remainingVideoUrls=${remainingVideoUrls.length}`)
+      }
       return {
         ...note,
         videoBuffer: buffer,
@@ -199,6 +256,10 @@ async function prepareNoteVideo(ctx: Context, note: XhsNote, config: Config): Pr
       ? await createTempVideoFile(buffer, mimeType)
       : `base64://${buffer.toString('base64')}`
 
+    if (config.loggerinfo) {
+      logger.info(`use downloaded video ${mode}: src=${mode === 'file' ? replacement : `base64://${formatBytes(buffer.length)} raw`}, remainingVideoUrls=${remainingVideoUrls.length}`)
+    }
+
     return {
       ...note,
       videoBuffer: undefined,
@@ -211,6 +272,53 @@ async function prepareNoteVideo(ctx: Context, note: XhsNote, config: Config): Pr
   }
 }
 
+async function getRemoteVideoSize(ctx: Context, url: string, config: Config): Promise<number | undefined> {
+  const maxSendSizeBytes = getSizeLimitBytes(config.maxVideoSendSizeMB, 100)
+  if (!maxSendSizeBytes) return undefined
+
+  try {
+    if (config.loggerinfo) logger.info(`check remote video size start: url=${url}`)
+    const headers = await ctx.http.head(url, { timeout: config.timeout * 1000 })
+    const contentLength = headers.get('content-length')
+    if (!contentLength) {
+      if (config.loggerinfo) logger.info('check remote video size skipped: missing content-length')
+      return undefined
+    }
+
+    const size = Number.parseInt(contentLength, 10)
+    if (!Number.isFinite(size) || size < 0) {
+      if (config.loggerinfo) logger.info(`check remote video size skipped: invalid content-length=${contentLength}`)
+      return undefined
+    }
+
+    if (config.loggerinfo) logger.info(`check remote video size success: size=${formatBytes(size)}, max=${formatBytes(maxSendSizeBytes)}`)
+    return size
+  } catch (error) {
+    if (config.loggerinfo) logger.info(`check remote video size failed: ${error instanceof Error ? error.message : String(error)}`)
+    return undefined
+  }
+}
+
+function removeNoteVideos(note: XhsNote): XhsNote {
+  return {
+    ...note,
+    videoBuffer: undefined,
+    videoUrls: [],
+    videoSkippedMessage: VIDEO_TOO_LARGE_MESSAGE,
+  }
+}
+
+function getSizeLimitBytes(sizeMB: number | undefined, fallbackMB: number): number {
+  const normalized = sizeMB ?? fallbackMB
+  return normalized > 0 ? normalized * 1024 * 1024 : 0
+}
+
+function formatBytes(bytes: number): string {
+  if (bytes < 1024) return `${bytes} B`
+  if (bytes < 1024 * 1024) return `${(bytes / 1024).toFixed(1)} KB`
+  return `${(bytes / 1024 / 1024).toFixed(1)} MB`
+}
+
 async function createTempVideoFile(buffer: Buffer, mimeType: string): Promise<string> {
   const fileName = `xhs-video-${Date.now()}-${Math.random().toString(16).slice(2)}${getVideoFileExtension(mimeType)}`
   const filePath = path.join(os.tmpdir(), fileName)

package/src/parser.ts

@@ -30,6 +30,7 @@ export interface XhsNote {
   videoUrls: string[]
   videoBuffer?: Buffer
   videoMimeType?: string
+  videoSkippedMessage?: string
 }
 
 const URL_BOUNDARY = '[^\\s"\'<>\\\\^`{|}，。；！？、【】《》]+'
@@ -123,6 +124,13 @@ export function buildXhsMessages(note: XhsNote, config: XhsConfigLike, session?:
     }
   }
 
+  if (note.videoSkippedMessage) {
+    messages.push(h('message', {
+      userId: session?.userId,
+      nickname: session?.author?.nickname || session?.username,
+    }, h.text(note.videoSkippedMessage)))
+  }
+
   return messages
 }